DakshBhesdadiya
AI EngineerMulti agent systems, large language models, and the automation that puts them to work.
- Python
- FastAPI
- LangChain
- LangGraph
- LangSmith
- CrewAI
- LlamaIndex
- RAG
- MCP
- LiveKit
- HuggingFace
- Ollama
- Anthropic
- Cohere
- Pinecone
- FAISS
- ChromaDB
- pgvector
- PostgreSQL
- SQLAlchemy
- Redis
- Supabase
- Docker
- AWS Bedrock
- n8n
- Make
- Transformers
- scikit-learn
- Pandas
- NumPy
Wiring language models into systems that actually finish the job. Agents that book the appointment, screen the candidate and run the interview, then hand a human only the part that needs one.
01Daksh Bhesdadiya is an AI engineer working in multi agent systems, large language models and generative AI. He builds the part of a product that thinks: agents that read a resume, route a call, hold a conversation, and then do something about it.
02Most of it is agentic rather than conversational. Not a chatbot bolted onto a form, but tools, memory and routing wired together with LangChain and LangGraph, deployed on AWS, and judged on whether the workflow finishes without a person stepping in.
Selected work
Four builds, all live.
Open any of them. Nothing here is a mockup.
- Dental AI Agent2025An appointment system for dental clinics that answers the phone. Voice and chat agents handle booking, rescheduling and reminders, with smart doctor routing and conflict free slotting behind them.LiveKitLangGraphFastAPIPostgreSQL
- Hiring Agent2025A hiring platform that does the screening. Resume parsing, candidate matching, scheduling and live AI interviews, with scorecards and analytics at the end of it. Manual screening down by roughly 80 percent.LangChainLLMsFastAPIMulti tenant
- Interview Coach2025Interview preparation that tells you the truth. ATS benchmarking scores a resume against the role, then simulated interviews give real time feedback and track whether the gaps are actually closing.LangChainRAGATS scoringAnalytics
- Voice Scheduling2025Calendar scheduling by speaking. A voice agent takes the request, resolves the conflicts and writes the event, with no form anywhere in the way.LiveKitMCPPythonCalendar APIs
What Daksh does
Agents that finish the job.
Multi agent systems
LangGraph, CrewAI, tool routing.
LLM applications
RAG, MCP, fine tuning, evals.
Voice agents
LiveKit, real time call handling.
Backend engineering
Python, FastAPI, PostgreSQL.
Workflow automation
n8n and Make, wired to real systems.
Deployment
AWS Bedrock, EC2, Docker, Nginx.
Daksh Bhesdadiya is an AI engineer at Kavion Solutions, working in multi agent systems, large language models and generative AI. He holds a BE in Information Technology from Government Engineering College Modasa and has built production AI platforms covering hiring automation, interview preparation and voice driven appointment booking.
He builds agentic AI systems: the tools, memory and routing that let a language model do a job end to end rather than just answer a question. That covers voice and chat agents, retrieval pipelines, workflow automation, and the Python and FastAPI backends underneath them.
LangChain, LangGraph, LangSmith and CrewAI for agents. LiveKit for voice. RAG and MCP for grounding models in real data, with Pinecone, FAISS, ChromaDB and pgvector as the vector stores. Python, FastAPI, SQLAlchemy and PostgreSQL on the backend, deployed with Docker, Nginx, AWS Bedrock and EC2.
A hiring platform that automates resume parsing, candidate matching and live AI interviews, which cut manual screening by roughly 80 percent. An interview preparation agent using ATS benchmarking and simulated interviews. A dental appointment system with voice and chat agents handling booking and patient engagement. Plus a voice driven calendar scheduling agent.
Email hello@kavionsolutions.in for anything involving Kavion Solutions work, or connect on LinkedIn and GitHub. Replies usually come within one business day.
The rest of the team